Artificial Intelligence in Medicine 55 (2012) 127–135 Contents lists available at SciVerse ScienceDirect Artificial Intelligence in Medicine journal homepage: www.elsevier.com/locate/aiim An automated methodology for levodopa-induced dyskinesia: Assessment based on gyroscope and accelerometer signals Markos G. Tsipouras a , Alexandros T. Tzallas a , George Rigas a , Sofia Tsouli b , Dimitrios I. Fotiadis a,∗ , Spiros Konitsiotis b a Unit of Medical Technology and Intelligent Information Systems, Dept. of Materials Science and Engineering, University of Ioannina, GR 45110 Ioannina, Greece b Dept. of Neurology, Medical School, University of Ioannina, GR 45110 Ioannina, Greece article info Article history: Received 30 November 2010 Received in revised form 23 February 2012 Accepted 4 March 2012 Keywords: Wearable devices Accelerometers Gyroscopes Classification Levodopa-induced dyskinesia assessment Parkinson’s disease abstract Objective: In this study, a methodology is presented for an automated levodopa-induced dyskinesia (LID) assessment in patients suffering from Parkinson’s disease (PD) under real-life conditions. Methods and Material: The methodology is based on the analysis of signals recorded from several accelerometers and gyroscopes, which are placed on the subjects’ body while they were performing a series of standardised motor tasks as well as voluntary movements. Sixteen subjects were enrolled in the study. The recordings were analysed in order to extract several features and, based on these features, a classification technique was used for LID assessment, i.e. detection of LID symptoms and classification of their severity. Results: The results were compared with the clinical annotation of the signals, provided by two expert neurologists. The analysis was performed related to the number and topology of sensors used; several different experimental settings were evaluated while a 10-fold stratified cross validation technique was employed in all cases. Moreover, several different classification techniques were examined. The ability of the methodology to be generalised was also evaluated using leave-one-patient-out cross validation. The sensitivity and positive predictive values (average for all LID severities) were 80.35% and 76.84%, respectively. Conclusions: The proposed methodology can be applied in real-life conditions since it can perform LID assessment in recordings which include various PD symptoms (such as tremor, dyskinesia and freezing of gait) of several motor tasks and random voluntary movements. © 2012 Elsevier B.V. All rights reserved. 1. Introduction Parkinson’s disease (PD) is a neurodegenerative disorder of the central nervous system that is manifested clinically by tremors, bradykinesia, rigidity, flexed posture, postural instability and freez- ing of gait [1]. The number of persons with PD over 50 years old in western Europe’s five most populous nations rose to 4.6 million in 2005, and this figure is expected to reach 9.3 million by 2030 [2]. The identification of the main cause of PD, i.e. the loss of brain cells that produce dopamine which helps coordinate and control muscular activity, led to the introduction of levodopa as a treat- ment. Levodopa is highly effective in reducing the symptoms of the disease and remains the standard drug for patients suffering from PD [3]. However, long-term PD treatment using levodopa is often ∗ Corresponding author at: Unit of Medical Technology and Intelligent Information Systems, Dept. of Materials Science and Engineering, University of Ioannina, PO Box 1186, GR 451 10 Ioannina, Greece. Tel.: +30 26510 08803; fax: +30 26510 07092. E-mail address: fotiadis@cc.uoi.gr (D.I. Fotiadis). complicated by significantly disabling fluctuations and dyskinesias, referred to as levodopa-induced dyskinesia (LID). LID is manifested as jerky, dance-like movements of body parts such as limbs (arms, legs), torso and head [4]. LIDs appear gradually and their severity increases progressively, and once established LIDs are difficult to treat. Therefore, efforts must be made in the direction of preven- tive strategies, which mainly focus on the optimal adjustment of the levodopa dosage. Prolonging the establishment of LID and minimis- ing its symptoms through optimal adjustment of levodopa dosage can only be achieved through long-term assessment of LID and its severity in PD patients. In addition, effective characterisation and quantification of LID improves the understanding of its pathophys- iological mechanisms, and helps towards treatment evaluation. Clinical methods currently used for LID assessment lack objec- tivity and cannot be used for long-term monitoring [5]. To overcome the limitations of the short-term and subjective assess- ments of LID, and to gain insight into the pathophysiology of LID episodes, several computer-based methods have been pre- sented in the literature [6–16]. These methods are based on the analysis of signals obtained using quantitative instrumental 0933-3657/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.artmed.2012.03.003